Rumor Spreading Modeling: Profusion versus Scarcity Martine - - PowerPoint PPT Presentation

Context Spatial model of rumor spreading Simulation Conclusion

Rumor Spreading Modeling: Profusion versus Scarcity

Martine Collard*, Philippe Collard**, Laurent Brison***, Erick Stattner* * LAMIA Laboratory, University of the French West Indies, France ** I3S Laboratory, University of Nice-Sophia Antipolis, France *** Lab-STIC Laboratory, Telecom Bretagne, France DYNO Workshop - ASONAM 2015

DYNO Workshop - ASONAM 2015 1 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Dissemination of rumors

Information: news, opinion, disease, virus, rumors How to model the propagation? Analogies among different kinds of information Particular emphasis on epidemical models in the literature Compartment models, Meta-population models, Network-based models Formal representation, Agent based simulations and Real data analysis Word of mouth rumors and Online rumors

DYNO Workshop - ASONAM 2015 2 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Outline

1

Context

2

Spatial model of rumor spreading

3

Simulation

4

Conclusion

DYNO Workshop - ASONAM 2015 3 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Outline

1

Context

2

Spatial model of rumor spreading

3

Simulation

4

Conclusion

DYNO Workshop - ASONAM 2015 4 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Context

Compartment models

Rumors studied for years originally in economics, psychology and social sciences (Knapp 1944, Rosnow & Fine 1976) Epidemiological mathematics and stochastic solutions (Kermack & Kendrick 1927) Agent-based and data driven simulations SIR: Three compartments (Susceptible Infected Recovered)

S I R

α β

DYNO Workshop - ASONAM 2015 5 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Context

Daley Kendall model

Rumor propagation is highly specific Currently circulating story or report of uncertain or doubtful truth (Oxford dictionnary) A kind of contagion process Multi-dimensional process driven mainly by socio-psychological elements DK Model (Daley & Kendall, 2006): Three compartments (Ignorant, Spreader and Stifler) Ignorant (I) Spreader (Sp) Stiffler (St)

p p'

DK: principle of novelty for an individual likely to tell the story, there is a reluctance to tell stale news

DYNO Workshop - ASONAM 2015 6 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Context

SIR and DK Assumptions

Assumptions in SIR and DK The population is fully mixed Individuals with whom a susceptible individual has contacts are chosen at random into the whole population All individuals have approximately the same number of contacts in the same period of time The transition from one state to another one relies on a same probability for every individual

DYNO Workshop - ASONAM 2015 7 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Context

Rumor specific properties

On the global scale Longness since it takes a relative long time for Spreaders to tell the story so that the rumor starts (Zhao al. 2012) Slowness since the propagation starts slowly and the information spreads in a short time (Kawachi al. 2008) Incompleteness since the infection does not reach the whole population (Daley and Kendall 2006) Sparseness since individual neighborhoods are not densely populated by Spreaders (Kawachi al. 2008) Rosnow and Fine (1976) identified the feature of scarcity as a key dimension for rumor spreading: Rumours arise when information is scarce

DYNO Workshop - ASONAM 2015 8 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Context

This work

Word of mouth rumors very difficult to follow, they do not generate data as online rumors that propagate on social media Investigation of the profusion/scarcity property of the rumor New perspective on the context of individuals likely to tell the story themselves once they know it A spatiotemporal model of rumor spreading What is the most realistic between the two antagonistic properties profusion and scarcity to disseminate a rumor?

DYNO Workshop - ASONAM 2015 9 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Outline

1

Context

2

Spatial model of rumor spreading

3

Simulation

4

Conclusion

DYNO Workshop - ASONAM 2015 10 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Spatial model of rumor spreading

ODS Model

A rumor is transmitted by word of mouth from individuals to individuals Spatiality, contact, social environment and psychological context are cornerstones of the phenomena ODS model relies on physical contact and mobility Each individual has a location in a world represented by a grid composed of cells His/Her neighborhood is composed by the others around him/her Individuals are mobile, they create new social contacts when they move The diffusion is relying on the induced social network and on individual spreading behaviour With spatiality, mobility, dynamicity and socio-psychological aspect, the goal is to better fit the framework to reality

DYNO Workshop - ASONAM 2015 11 / 23

Context Spatial model of rumor spreading Simulation Conclusion

ODS model

Compartments

Open-minded agents are the individuals who have not yet heard the rumor, and, consequently, are susceptible to becoming informed Disseminators are active individuals that are spreading the rumor Stiflers are individuals that have got the rumor but are no longer spreading it The total population size N = O + D + S

DYNO Workshop - ASONAM 2015 12 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Spatial model of rumor spreading

ODS vs SIR and DK

The aim is to model the spreading of a rumor, thus with the spatial dimension Contacts between individuals are not chosen at random and they occur between neighbors only Agents are moving, at each time step, the number of contacts for an individual is not a constant The probability that a D-individual transmits the information to an O-individual upon contact depends on each individual and varies over

• time. It is referred as βOD

k (t)

DYNO Workshop - ASONAM 2015 13 / 23

Context Spatial model of rumor spreading Simulation Conclusion

ODS Model

Transitions

The potential receiver, and not the transmitter, decides whether or not he will become itself a transmitter: Crucial difference with infectious disease spreading An O-individual ak becomes himself a D-individual according to a function of the rate r D

k of D-individuals in his neighborhood.

Two alternative solutions:

1

ODSp driven by the profusion of information

2

ODSs driven by the scarcity of information according to the number of D-individuals in the vicinity

γ

O D

Open minded Disseminator

S

Stiffler β

depends on Disseminator scarcity/profusion in neighborhood depends on time being a Disseminator

DYNO Workshop - ASONAM 2015 14 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Outline

1

Context

2

Spatial model of rumor spreading

3

Simulation

4

Conclusion

DYNO Workshop - ASONAM 2015 15 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Simulation

ODS Algorithm

• 1. t ← 0
• 2. Initialize the parameter DPeriod {γ ←

1 DP eriod}

• 3. Initialize the population size to N
• 4. Create N agents

{each agent have a state variable in {O, D, S}}

• 5. Place at random the N agents on the grid

{each agent have a position in the 2-D space} {each agent have a heading which indicates the direction he is facing}

• 6. Set all the agents O except one which is D
• 7. while 9 one D agent do

8.

Call OtoD

9.

Call DtoS

10.

Call walk {ask all agents to move}

11.

t ← t + 1

• 12. end while

Ensure: 6 9 D agent

DYNO Workshop - ASONAM 2015 16 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Simulation

Transmissibility potential

To report if the invasion will succeed or not: Reproductive ratio R0 = β×N

γ

that is the number of secondary infections that result from a single Infected individual in a fully Susceptible population ODSs results are consistent with the longness feature

Figure : Probability for a rumor to occur plotted against Dperiod

DYNO Workshop - ASONAM 2015 17 / 23

Context Spatial model of rumor spreading Simulation Conclusion

ODS model

Rumor curve

Rumor curve as an epidemic curve: ODSs process is starting much slower and has a smaller amplitude than in case of profusion

(a) (b)

Figure : Rumor curve (Dperiod = 10), profusion (a) scarcity (b)

DYNO Workshop - ASONAM 2015 18 / 23

Context Spatial model of rumor spreading Simulation Conclusion

ODS model

Number of individuals in each compartment

At the end of the process, with ODSs there remain individuals that are not Stiflers Consistent with the incompleteness feature

(a) (b)

Figure : Proportion of individuals in each compartment as a function of time (Dperiod = 10), profusion (a), scarcity (b)

DYNO Workshop - ASONAM 2015 19 / 23

Context Spatial model of rumor spreading Simulation Conclusion

ODS model

Spatial distribution of compartments

D-aggregation index agD(t) =

1

|D| ∑k∈D pD

k (t)

where pD

k (t) can be interpreted as the profusion of disseminators around an

individual ak. Low values of agD(t) means that D-individuals will be spread over the world High values correspond to configurations with more homogeneous patterns

• f D-individuals

D-aggregation index evolution: blue curve

DYNO Workshop - ASONAM 2015 20 / 23

Context Spatial model of rumor spreading Simulation Conclusion

ODS model

Spatial distribution of compartments

Profusion

(a) t = 70 (b) t = 70 (c) t = 70

Figure : Spatial distribution of compartments (Dperiod = 10) with profusion and O-persons (a), D-persons (b), S-persons (c)

DYNO Workshop - ASONAM 2015 21 / 23

Context Spatial model of rumor spreading Simulation Conclusion

ODS model

Spatial distribution of compartments

Scarcity

(a) t = 120 (b) t = 120 (c) t = 120

Figure : Spatial distribution of compartments (Dperiod = 10) with scarcity and O-persons (a), D-persons (b), S-persons (c)

Consistent with sparseness

DYNO Workshop - ASONAM 2015 22 / 23

Context Spatial model of rumor spreading Simulation Conclusion

Conclusion

We have proposed a model of rumor diffusion that integrates the local context of individuals making the phenomenon Agent-based simulations have shown that scarcity induces characteristic features of a rumor identified as longness, slowness, incompletness and sparceness Next step will consider an explicit social network like a scale free network

DYNO Workshop - ASONAM 2015 23 / 23